Shahina Begum

Zero-day network attacks pose a significant threat because their unknown signatures evade traditional detection mechanisms. This research develops an AI-enhanced intrusion detection system that aims to detect such attacks while providing interpretable outputs for security analysts. Four machine-learning models are evaluated under strict zero-day conditions using two benchmark datasets. SHAP and LIME are applied to produce instance-level explanations, and a formal stability assessment is conducted to determine their reliability. Experimental results show that the hybrid model combining anomaly-based detection with deep learning achieves the highest zero-day Recall, with statistically significant advantages over individual models in detecting previously unseen attacks, while the standalone LSTM achieves the strongest overall balance between Precision and Recall. The generated explanations consistently reveal security-relevant features, and stability analysis confirms their robustness across conditions. The study demonstrates that integrating deep learning with stable explainable AI offers a practical and trustworthy solution for zero-day intrusion detection, contributing validated evidence to an area where explanation reliability is rarely examined.

Artificial Intelligence has revolutionised industries, enhancing productivity and efficiency across different sectors. However, its deployment in life-critical domains such as road and air traffic safety, demands trustworthiness in the sustainable development of AI technology. The challenge is to develop AI solutions with long-term sustainability in mind, ensuring they remain relevant and reliable to the users. This paper highlights the role of eXplainable AI (XAI) and transparency in ensuring trustworthiness and fostering sustainability in AI solutions. XAI techniques, which can often be interpretable models and post-hoc explanations, are particularly vital in domains where AI decisions have significant implications, such as predicting machinery failures in manufacturing. Despite its potential, achieving explainability remains challenging due to trade-offs between model complexity and interpretability, the absence of universal standards, and diverse societal expectations. The paper outlines key steps to ensure explainability and transparency in reliable AI development supported by relevant use cases.

Multi-agent systems (MAS) based on large language models (LLMs) are on the rise due to their improved capabilities compared to single agents. In particular, among the emerging MAS for software development, those establishing automated agentic workflows contribute to the idea that software development is an essentially solved task. This study aims at critically assessing five MAS for software development that cover multiple phases and support automated workflow: MetaGPT, ChatDev, AgileCoder, HyperAgent and AgentLite. Here, 14 research questions are identified, specifically referring to architectural-, technical-, operational-, human interaction- and responsible AI-aspects. The findings from the study can help interested researchers and practitioners to understand which design choices bring advantages for particular aspects and where, on the contrary, existing solutions are suboptimal and what could be suggestions for addressing the gaps. Due to its research question-oriented layout, this paper is ideal as a reference which, for a particular aspect, facilitates grasping the overview of main design choices, capabilities, limitation and reconciliation approaches.

Explainability is crucial for establishing user trust in Artificial Intelligence (AI), particularly within safety-critical domains such as Air Traffic Management (ATM) and Air Traffic Control (ATC). This study empirically investigates the effects of Explainable AI (XAI), specifically HeatMap-based visual explanations, on cognitive workload, user acceptance, and intention to use AI-driven decision-support systems among Air Traffic Control Officers (ATCOs). Despite significant theoretical advancements in the broader XAI domain, empirical evidence addressing the specific impact of visual explanations on human-AI interactions in safety-critical environments like ATC remains limited. To address these critical gaps, an experimental comparison was conducted between explainable (HeatMap) and non-explainable (BlackBox) AI conditions, involving two user groups: expert and student ATCOs. Both objective neurophysiological measures (Electroencephalography) and subjective questionnaires were employed to capture comprehensive user responses. Key findings revealed that the presence of visual explanations significantly reduced cognitive workload and enhanced users’ willingness to adopt the AI system, regardless of participants’ level of expertise. However, explicit perceptions of AI’s impact on work performance were predominantly influenced by expertise, with less experienced controllers reporting a greater perceived impact than their expert counterparts. By combining objective neurometrics with subjective user assessments, this research advances methodological rigor in evaluating human-AI interactions and highlights the importance of tailored, user-centric explanations. These findings directly contribute to practical guidelines for designing cognitively compatible and trustworthy AI tools in ATC, providing nuanced insights for targeted training and deployment strategies based on user expertise.